Since the 1980s, when dengue was reintroduced in Brazil, outbreaks and epidemics caused by different arbovirus strains transmitted by vector mosquitoes such as Aedes aegypti (A.e) have been an annual occurrence . Since 2010, it has been estimated that there have been over a million cases of dengue per year, leading to hundreds of deaths . Since 2015, with the introduction of the Zika and Chikungunya viruses in the country, cases have been reported of children with congenital Zika syndrome or chronic sequelae . The similarity of symptoms, cross-reactivity, co-circulation and overlap of infections by different arbovirus species and strains make the differential diagnosis of these diseases difficult . The main strategy for reducing the prevalence of these diseases has been vector control with larvicide application in water tanks and spatial nebulization with insecticides widely used in the control of adult specimens , which has contributed to the selection of resistant insect strains . The development of vaccines and alternative control methods using transgenic bacteria and mosquitoes has also been reported in the literature [5, 6].
The engagement of the population in preventive actions is one of the main objectives of vector control policies, given that most of the breeding sites are located in households and on vacant land and abandoned houses [7, 8]. This engagement has been stimulated and supported by mass media advertising, the use of breeding reporting applications (eg Zikazero)  and environmental education by home-visiting performed by health workers [7, 10]. Unfortunately, in urban centres in Brazil, these visits often do not occur because residents are not in their homes or they do not allow health workers to enter for inspection and guidance on vector control for fear of violence . In this context, the implementation of interventions in schools using digital platforms for mobile devices can be a powerful tool for the promotion of engagement by the population in preventive actions for arboviruses, and this can be directed by entomological surveillance services.
The World Health Organization has recommended grounding educational interventions in behaviour change theories [12, 13], considering the growth of evidence for their effectiveness at individual, community and population levels [14, 15]. The Health Belief Model (HBM), for example, suggests that decision-making depends on the perception of susceptibility, disease severity, benefits and barriers associated with behaviour [12, 16]. Social Cognitive Theory (SCT) describes three main factors that affect a person's likelihood of changing health behaviour: self-efficacy, goals, and outcome expectations [12, 17]. Even small changes in human behaviours can have substantial effects on population health outcomes .
Building on these behaviour change theories, the aim of this study was to evaluate the behavioural change of high school students and teachers who participated in a virtual educational intervention. This is the third stage of the project, “Impact of mobile learning on prevention and management of complications caused by arbovirus (Zika, Dengue, Chikungunya) – ZikaMob,” funded by the British Council and the Government of Paraíba State, in Northeastern Brazil.
Design and scenario
This is a school-based intervention in which a self-reported questionnaire was used before and after the intervention to assess environmental risk factors and sociodemographic variables, and to measure attitudes and behaviours. The theoretical and methodological foundations of the research, the software development process and the validation of the data collection instrument were described previously [18,19,20].
Campina Grande is the second largest city in the state of Paraíba, housing an estimated population of 409,731 in 2019, with 15,152 students and 1,732 teachers registered in high schools (IBGE). The city is in the semiarid northeastern region and experiences prolonged dry periods; for this reason, the state periodically establishes water rationing, ranging from two to four days a week, causing the population to use water tanks or buckets for water storage. From 2018, with the transposition of waters of the São Francisco River, there was a recharge of water in supply dams with a concomitant suspension of the water supply rotation. The city had high rates of arbovirus vector mosquito infestation, with an associated high risk for arbovirus outbreaks and epidemics .
The intervention consisted of a competition between high schools in order to fulfill the largest number of arbovirus prevention educational activities (“missions”). A virtual platform called ZikaMob was established to allow the follow-up of the execution of these “missions” and the inclusion of audio-visual content published on social networks. To participate in the project, students were registered on the platform through an invitation sent to their personal email and following their guardians’ signing of the consent form. Each participant had an individual password, could access the platform from any mobile device and were able to track activities virtually or via the Zikamob Facebook page.
The Zikamob platform is a virtual learning environment, like Moodle or Google Classroom, allowing the inclusion of didactic material, quizzes, questionnaires, and videos. Unlike existing platforms, on Zikamob participants could include the Facebook URLs of their videos to prove that they had completed the mission. By including their social media posts or answering the questionnaires on the Zikamob platform, the student received points. The sum of students' points was used to compose the total points for each school. The individual and school points were counted on the platform as well as likes, shares and interactions of their social media posts.
Every two weeks, participants were given a mission and had to produce a creative video to prove their accomplishment and post it on Facebook. In all, five video missions and two questionnaires were completed during a three-month period. The first mission was to make a video inviting people to follow the Zikamob Facebook page (https://www.facebook.com/zikamob.uepb.3) and to engage in the activities. The second mission was to learn how to inspect the home, identify mosquito-borne arbovirus outbreaks, and how to properly eliminate them. Students had to watch a report with a health worker that explained how to complete these procedures. The third mission was to learn how to sort recyclables and donate to waste pickers, using an application called CATAKI that lets users locate waste pickers in order to match the delivery of recyclable material. In Campina Grande, like most Brazilian cities, there is no selective collection service and the trash found in the streets, backyards and open areas serves as fertile breeding areas for mosquitoes. The fourth mission was to learn how to screen windows and drains to prevent mosquitoes from entering homes. The final mission asked participants to make clean-up efforts in their backyards or vacant lots near their homes.
The ZikaMob project's Facebook page allowed the publication of calendars, warnings, and mission guidance. In schools, each class selected the top three videos, and these received extra points. Teachers were responsible for choosing the top three videos from the school for each of the missions, which also received extra points. A video representing each school in each of the five missions was posted on the ZikaMob Facebook page for the public to vote on through the awarding of likes. The three videos with the most likes for each mission were nominated for awards, for which the winners received cell phones and other awards. The observation of people taking preventive action is one of the factors that can promote desired behaviour change, according to the Social Cognitive Theory .
The project was initiated after approval by the Research Ethics Committee of the State University of Paraíba (Protocol CAAE 67429517.5.0000.5187) and due consent of the participants’ parents or guardians was given through the signing of the informed consent form. The registration of teachers and students in the ZikaMob platform was carried out from April to June, and the intervention during the months of July to September 2019.
The assessment of perceptions and behaviours before and after the intervention was performed by applying a self-reported questionnaire, containing questions with binary answers of the "yes" and "no” type. The questionnaire was organized in different sections that contained questions about socio demographic aspects, environmental, psychosocial, and behavioural factors, related to the prevention of mosquito-borne arboviruses. Each response was classified for one point if it was a preventive factor, or zero points otherwise. For example, each participant was asked whether they had covered water tanks at home. Having uncovered tanks is a risk factor because it favours the proliferation of mosquitoes; therefore, this response was classified as a risk, receiving zero points. The sum of points for each section defined a score, so that it was possible to evaluate a set of answers together. The more points, the more preventive behaviours people performed at the time the questionnaire was applied. The questionnaires were made available on the ZikaMob platform and answered by participants through their mobile devices. The validation of the questionnaire was described in the previous works [18, 19].
The dependent variables (D) correspond to the target behaviours of educational intervention, subdivided into two groups. To compose the Target Behaviour Score (D), participants were asked whether their water reservoirs, trash cans and capped drains were covered to avoid mosquitoes’ proliferation; whether their families survey reservoirs at least once a month, clean their water tanks, separate solid waste for recycling, inspect potted plants, close windows at dawn and dusk, and clean vacant lots. Each preventative behaviour performed received one point, so the score could vary from zero to eleven points. The Breeding Identification and Elimination Score (DC), ranging from zero to four points, was related to the behaviours reported by participants when identifying a mosquito breeding site. The participant had to know how to eliminate it; have learnt to throw water on sunny ground to eliminate the larvae; to wash containers; inspect other places in the residence for more breeding sites; to alert neighbours to the danger and to notify the Environmental Surveillance service about a prospective danger of infestation.
The Independent variables were grouped into subgroups. Sociodemographic variables consisted of gender, age, and role (student or teacher). Environmental and household risk were assessed through indicators such as: access to garbage collection services; running water and whether water was lacking two or more days a week; whether the household had a yard, plants, cistern, water tank or other water reservoirs that increase the risk of breeding; whether the residence was a single-storey house or building; and whether it was owned or rented. The higher the home or environmental risk in relation to having mosquito breeding sites, the higher the Risk Score (R) ranging from zero to 14 points.
The Facilitator Behaviour Score (F) refers to the fact that the participant already has some practices that may favour the accomplishment of the target behaviour, such as doing housework or taking care of potted plants or gardens. In order to compose the Perception of Prevalence Score (P), ranging from 0 to 5 points, two levels of knowledge were assessed: 1) that of the participant (and, by extension, their family) about mosquito-borne arboviruses, and 2) their self-perception about diagnosing these diseases. Participants were asked whether they or their family members had had Zika, Dengue or Chikungunya; whether it is possible to get dengue more than once in their life; whether all mosquitoes transmit dengue fever and whether a dengue vaccine has already been developed.
Some constructs of behaviour change theories were also assessed by answering individual questions or adding points to the overall score. These are all grouped under the T score but can be broken down as follows: The Self-Efficacy Score assesses the self-efficacy and collective efficacy constructs that reveal whether a person believes they can perform the behaviour and change the behaviour of their family and neighbours. The constructs of the Health Belief Model (Health Belief Score) are concerned with the perception of the susceptibility, severity, barriers, and benefits associated with behaviour change. Each positive response meant having a certain belief or attitude that could favour preventative behaviour.
Sample and Statistical Analysis
The study population was composed of all students and teachers of Campina Grande High Schools who agreed to participate in the project and signed the free and informed consent form. Out of a total of 3,681 students invited to participate voluntarily in this research, 883 (24%) students were included in this study because they answered at least one of the questionnaires used to assess attitudes, beliefs, and behaviours. Of this total, 227 participants answered the questionnaire before and after the intervention, and statistical tests were performed for paired samples to assess the change in behaviour of this group (Study A). An independent sample analysis was performed with a total of 626 participants, 364 of whom answered the questionnaire only before the intervention (Q1) and 262 who answered the questionnaire after the intervention (Q2) (Study B).
Descriptive statistics were used to describe the population profile and the frequency of each questionnaire response before and after the intervention. The normality test showed that the scores and age did not follow normal distribution, so only nonparametric tests were used for inferential analysis. Pearson's chi-square tests were used; Wilcoxon tests were also performed where paired samples were available, and Kruskal Wallis tests where there were independent samples (SIEGEL, 2006), both using the significance level of 5% (p-value <0.05). The reliability analysis of the instrument was performed using Cronbach's alpha, a test scorereliability coefficient for categorical variables. The analyses were performed with the aid of the R statistical software [21, 22].
In a second iteration of analysis, multivariate analysis techniques were used, adjusting the Principal Component Analysis (PCA), whose eigenvalues were greater than one (λ> 1), as suggested by Kaiser (1960), in order to identify a smaller number of variables: uncorrelated alternatives that somehow summarize the main information of the original variables. Subsequently, these main components were presented in Biplot graphs for individuals and variables with their respective confidence ellipses (with 95% reliability). Biplot is a method that represents two-dimensional multivariate data, where each observation is represented by the pair of scores of the first two main components, representing each group in their respective confidence ellipses. The PCA aimed to compare the patterns of these ellipses for the group of students and teachers, analysed before and after the intervention.